替换embedding模型

Dockerfile

@@ -39,7 +39,7 @@ RUN mkdir -p /app/public
 RUN mkdir -p /app/models
 
 # 下载sentence-transformers模型到models目录
-RUN python -c "from sentence_transformers import SentenceTransformer; model = SentenceTransformer('sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2'); model.save('/app/models/paraphrase-multilingual-MiniLM-L12-v2')"
+RUN python -c "from sentence_transformers import SentenceTransformer; model = SentenceTransformer('TaylorAI/gte-tiny'); model.save('/app/models/gte-tiny')"
 
 # 暴露端口
 EXPOSE 8001

Dockerfile.modelscope

@@ -40,7 +40,7 @@ RUN mkdir -p /app/public
 RUN mkdir -p /app/models
 
 # 从modelscope下载sentence-transformers模型到models目录
-#RUN python -c "from modelscope import snapshot_download; model_dir = snapshot_download('sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2'); import shutil; shutil.move(model_dir, '/app/models/paraphrase-multilingual-MiniLM-L12-v2')"
+#RUN python -c "from modelscope import snapshot_download; model_dir = snapshot_download('TaylorAI/gte-tiny'); import shutil; shutil.move(model_dir, '/app/models/gte-tiny')"
 
 # 复制应用代码
 COPY . .

embedding/manager.py

@@ -25,9 +25,9 @@ logger = logging.getLogger(__name__)
 class GlobalModelManager:
     """全局模型管理器"""
     
-    def __init__(self, model_name: str = 'sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2'):
+    def __init__(self, model_name: str = 'TaylorAI/gte-tiny'):
         self.model_name = model_name
-        self.local_model_path = "./models/paraphrase-multilingual-MiniLM-L12-v2"
+        self.local_model_path = "./models/gte-tiny"
         self._model: Optional[SentenceTransformer] = None
         self._lock = asyncio.Lock()
         self._load_time = 0
@@ -64,8 +64,7 @@ class GlobalModelManager:
                     None, 
                     lambda: SentenceTransformer(
                         model_path,
-                        device=self._device,
-                        truncate_dim=128
+                        device=self._device
                     )
                 )
                 
@@ -126,6 +125,6 @@ def get_model_manager() -> GlobalModelManager:
     """获取模型管理器实例"""
     global _model_manager
     if _model_manager is None:
-        model_name = os.getenv("SENTENCE_TRANSFORMER_MODEL", "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2")
+        model_name = os.getenv("SENTENCE_TRANSFORMER_MODEL", "TaylorAI/gte-tiny")
         _model_manager = GlobalModelManager(model_name)
     return _model_manager

utils/data_merger.py

@@ -198,7 +198,7 @@ def merge_embeddings_by_group(unique_id: str, group_name: str) -> Dict:
         dimensions = 0
         chunking_strategy = 'unknown'
         chunking_params = {}
-        model_path = 'sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2'
+        model_path = 'TaylorAI/gte-tiny'
         
         for filename_stem, embedding_path in sorted(embedding_files):
             try: